Apparatus and method for detecting entities in an image

Abstract

An apparatus and a method are provided for detecting entities in a numerical image, wherein the apparatus includes a computing unit configured for detecting, based on a histogram vector determined on the basis of gradient and partitioning information, the presence of at least one of the entities in the image, a signaling unit in signal communication with the computing unit, and configured for being activated when the computing unit detects the presence of at least one of the entities in the image, memory containing partitioning information, and configured for allowing access to the partitioning information on the basis of the gradient information, wherein each piece of partitioning information identifies at least one of the partitioning elements that allow the computing unit to quantize the gradient information.

The invention claimed is: 
     
       1. An apparatus adapted to detect entities in a numerical image comprising a plurality of points, comprising:
 computing means configured for:
 reading or determining at least one piece of gradient information relating to one of the points of the image, 
 generating a histogram vector on the basis of at least said gradient information and of partitioning elements with reference to which at least said gradient information is quantized, and 
 detecting, based on said histogram vector, the presence of at least one of said entities in the image, and 
 signaling means in signal communication with the computing means, and configured for being activated when said computing means detect the presence of at least one of said entities in the image, and 
 
 memory means containing at least a plurality of pieces of partitioning information, and configured for allowing access to at least one of said pieces of partitioning information on the basis of said gradient information, and wherein each one of said pieces of partitioning information identifies at least one of said partitioning elements allowing the quantization of said gradient information, so as to allow the generation of the histogram vector. 
 
     
     
       2. The apparatus according to  claim 1 , wherein the gradient information comprises the vertical G V (x, y) and horizontal G H (x, y) gradient values associated with said point of the image. 
     
     
       3. The apparatus according to  claim 2 , wherein the memory means contain also a plurality of pieces of modulus information comprising correction factors, wherein said memory means are also configured for allowing access to at least one of said pieces of modulus information on the basis of said gradient information, and wherein each correction factor allows computing the modulus of the gradient associated with said point of the image, by multiplying said correction factor by the sum of the vertical (G V (x, y)) and horizontal (G H (x, y)) gradient values. 
     
     
       4. The apparatus according to  claim 2 , wherein the memory means contain also a plurality of pieces of projection information, and are also configured for allowing access to at least one of said pieces of projection information on the basis of said gradient information, and wherein each piece of projection information comprises a first correction factor that allows computing the modulus of the projection of the gradient associated with said point of the image on the partitioning element obtained from the memory means on the basis of said gradient information, by multiplying said correction factor by the sum of the vertical (G V (x, y)) and horizontal (G H (x, y)) gradient values. 
     
     
       5. The apparatus according to  claim 4 , wherein each one of said pieces of partitioning information comprises also a second partitioning element, and wherein each piece of projection information comprises also a second correction factor that allows computing the modulus of the projection of the gradient associated with said point of the image on said second partitioning element, by multiplying said second correction factor by the sum of the vertical (G V (x, y)) and horizontal (G H (x, y)) gradient values. 
     
     
       6. The apparatus according to  claim 2 , comprising normalizing means configured for being inputted the vertical G V (x, y) and horizontal G H (x, y) gradient values and for outputting a pair of normalized values (G H (x, y)*, G V (x, y)*), which are used for accessing the memory means, and wherein said normalized values are generated by subjecting both the vertical and horizontal gradient values (G V (x, y), G H (x, y)) to the same number of binary shifts towards the most significant bit and to a truncation of the same number of least significant bits. 
     
     
       7. The method for detecting entities in a numerical image comprising a plurality of points, comprising
 a gradient determination phase, for reading or computing at least one piece of gradient information relating to one of the points of the image, 
 a histogram generation phase, for generating a histogram vector on the basis of at least said gradient information and of partitioning elements with reference to which at least said gradient information is quantized, 
 an entity detection phase, for detecting, based on said histogram vector, the presence of at least one of said entities in the image, 
 an entity presence signaling phase, wherein, if during the entity detection phase the presence of at least one entities is detected in the image, said presence will be signaled by means for signaling, 
 wherein during the histogram generation phase the partitioning elements are selected on the basis of at least said gradient information. 
 
     
     
       8. The method according to  claim 7 , wherein the gradient information comprises the vertical G V (x, y) and horizontal G H (x, y) gradient values associated with said point of the image. 
     
     
       9. The method according to  claim 8 , wherein, during the histogram generation phase, modulus information comprising correction factors is also selected along with said partitioning elements, and wherein each correction factor allows computing the modulus of the gradient associated with said point of the image, by multiplying said correction factor by the sum of the vertical (G V (x, y)) and horizontal (G H (x, y)) gradient values. 
     
     
       10. The method according to  claim 8 , wherein, during the histogram generation phase, projection information is also selected along with said partitioning elements, and wherein each piece of projection information comprises a first correction factor that allows computing the modulus of the projection of the gradient associated with said point of the image on the partitioning element associated with said gradient information, by multiplying said correction factor by the sum of the vertical (G V (x, y)) and horizontal (G H (x, y)) gradient values. 
     
     
       11. The method according to  claim 10 , wherein, during the histogram generation phase, a second partitioning element is also selected along with each one of said partitioning elements, and wherein each piece of projection information comprises a second correction factor that allows computing the modulus of the projection of the gradient associated with said point of the image on said second partitioning element, by multiplying said second correction factor by the sum of the vertical (G V (x, y)) and horizontal (G H (x, y)) gradient values. 
     
     
       12. The method according to  claim 7 , wherein, during the histogram generation phase, the vertical G V (x, y) and horizontal G H (x, y) gradient values are transformed into a pair of normalized values (G H (x, y)*, G V (x, y)*), which are used for selecting the partitioning elements, and wherein said normalized values are generated by subjecting both the vertical and horizontal gradient values (G V (x, y), G H (x, y)) to the same number of binary shifts towards the most significant bit and to a truncation of the same number of least significant bits. 
     
     
       13. A non-transitory computer readable medium comprising instructions that, when executed by a processor, perform the method according to  claim 7 .